Intel CEO Lip-Bu Tan: AI Hardware Shortage to Persist for Two to Three Years

On October 12, 2026, Intel CEO Lip-Bu Tan spoke at a panel discussion organized by DBS Bank in Singapore, where he addressed the state of Chinese AI models. According to Reuters, he said that frontier AI models from China have consistently low token costs and are spreading very quickly, adding that models such as DeepSeek, Kimi and GLM are already "good enough" in terms of performance.

The following day, IT之家 reported further remarks from the same event. Tan said that AI hardware will remain in tight supply for the next two to three years, citing continued constraints in the technology supply chain. He noted that as agentic AI applications become more widespread, demand is shifting toward CPUs rather than GPUs, and that the ratio of GPUs to CPUs in AI training has fallen from 8 to 1 down to 4 to 1. The report also mentioned that Intel's Data Center and AI business unit grew revenue 59% year over year in the second quarter, and that Tan had referenced the same ratio shift when discussing the company's Q2 earnings. The report added that his comments came as core components such as memory chips remain in short supply, with prices rising for months.

Both reports stem from Tan's remarks at the DBS Bank event in Singapore. The claims about Chinese model performance and token costs were relayed by Reuters, while the comments on hardware shortages and the GPU-to-CPU ratio came from IT之家. The available information stops at these assessments, with no specific timeline given for when the shortage might ease.